PREVIOUS ADVICE 
I have been having a running argument with my boss concerning the definition of mentorship, especially in regards to grants. I am an assistant research professor who has submitted an R-21 proposal. One of the comments from the reviewers was, "Mentorship or collaboration with an experienced investigator in the field would be helpful." My boss has experience in the cell-cycle field but not with herpes viruses; my project would encompass both fields. My boss believes that if I list two mentors--him and another scientist--that my grant wouldn't be funded as I will be torn between two mentors and nothing will get done. I do not agree with his position.
What is your definition of mentorship? Can a person have more than one mentor? And, especially with regards to a grant proposal, would mentioning two mentors be totally out of line?
Your question raises two distinct questions, one general and profound, the other specific and practical. I'll deal with the general and profound issue first.
Q. Can a scientist have more than one mentor?
The question of mentorship has gotten a lot of attention in recent years, as administrators, scientists, and policymakers have become aware of how important it is for a young scientist's development. One consequence of this attention is the evolution of a list of "best practices," or something like it. Although it is not, perhaps, a definitive statement, a good place to look for an informal treatment of the subject is Advisor, Teacher, Role Model, Friend published by the National Academies. It's available free online .
One point this publication makes--a point that's repeated ad nauseum in mentorship circles but is no less true and important for that--is that an advisor and a mentor are quite different beasts. The major difference is that mentoring relationships are personal, whereas advising relationships are, in their purest form, strictly professional. A mentor takes a personal interest in the person being mentored (some describe the object of mentoring as a "mentee" but to me that sounds too much like a breath mint), becoming a sort of senior friend; the person being mentored, in turn, emulates some aspect of the character, professional habits, or both, of the mentor.
So: Can you have more than one such person in your life? Can you have more than one friend? Of course you can, as a rule, though it depends on your individual temperament. You may find your needs filled by a single mentor; yet there is no harm in having multiple mentors, and there is much that is good, potentially.
Another key distinction between "mentor" and "advisor" is that mentorship acknowledges, in an important way, the individual needs of the person being mentored. A mentor is there for you to emulate, and to provide guidance when called upon, but proper mentors rarely impose a vision of how things ought to be. In assembling a life and a career for yourself you will invent a lot that is new, and you will, very likely, borrow even more, from a range of people. That is exactly as it should be. Which brings me to your more specific question:
Q. Should I have more than one mentor/senior collaborator on a grant proposal?
If this had been a mentored training award, like a K-08 or a K-22, I would have encouraged you to stick with a single mentor. In the context of the National Institutes of Health's (NIH's) mentored training awards, mentorship has a particular meaning: an expert in an area you wish to pursue, who, with you, is charged with setting out a compelling and original training program that meets your specific needs. Advice and instruction from others is very much encouraged, but the responsibility for your training rests, in most cases, with a single individual.
The R-21 is, not, however, a mentored award. It is a research grant. It is, admittedly, especially appropriate to early-career investigators. Its smaller size means that NIH isn't making too large a bet on someone with a minimal track record, which makes reviewers and study sections a little less cautious. The "exploratory" nature of the grant means that less preliminary data are expected. And because the term is shorter, the award is appropriate for independent scientists with short-term appointments.
Yet, for all of that, it is a research grant, and you are not your boss, you are the principal investigator. It is up to you to assemble a team that will allow you to get the work done. And it is up to you to convince NIH that you have what it takes. Unless you are personally conflicted, there is no conflict.
You asked about mentors, but your reviewer also used the word "collaborator," and that is the word that I prefer, because it more accurately describes the sort of relationship that is called for here. Any reasonably ambitious project is likely to stretch the PI beyond one's areas of expertise. The best way to make this work is to find a scientist whose expertise complements yours. Since you'll be the one calling the shots, what you need is a collaborator, not a mentor.
Why, then, did the reviewer add the word "mentor?" He might have intended to suggest a relationship where you would reap the benefit of your partner's expertise without giving them any of the money; R-21s, after all, are not large. Setting up such a relationship may be easier with a more established investigator than a younger one who, like you, needs cash. Still, I prefer to call it a collaboration, even if no money changes hands.
Covering all the bases like this is a wonderful formula for getting grants funded. Wherever there's a hole in your proposal, a place where your arguments are unconvincing and your knowledge is untested, bring in someone to help revise the proposal and assist you with the work. This will reassure the reviewers (and the study section) that (1) you are perceptive enough to recognize your limitations and resourceful enough to address them, and (2) you have a strong team in place. The former will help convince reviewers of your competence and potential as a PI; the latter will help convince them that the work will get done, and that is their main concern.
Importantly, your new collaborator will almost certainly be willing to read your proposal critically and suggest meaningful changes of substance and style. If there's anything more compelling to a reviewer than a proposal obviously informed by a deep and thorough knowledge of the nuances of a field, it's a proposal obviously informed by a deep and thorough knowledge of the nuances of two fields.
A close acquaintance took just this approach in writing an R-15 (AREA) grant, intended to fund research at less research-intensive schools. This scientist--a research-active faculty member at an excellent undergraduate institution who had never before applied for an NIH grant--prepared a strong conceptual draft, which was revised extensively on the advice of several collaborators, each of them expert in a certain, relevant area. In a study section that was simultaneously reviewing R-01 applications, this proposal received a priority score of 132.